diff --git a/ViennaRNAParser.cabal b/ViennaRNAParser.cabal --- a/ViennaRNAParser.cabal +++ b/ViennaRNAParser.cabal @@ -5,7 +5,7 @@ -- PVP summary: +-+------- breaking API changes -- | | +----- non-breaking API additions -- | | | +--- code changes with no API change -version: 1.2.1 +version: 1.2.2 synopsis: Libary for parsing ViennaRNA package output description: Currently contains parsers and datatypes for: RNAalifold, RNAdistance, RNAcofold, RNAfold, RNAplex, RNAup, RNAz. . @@ -30,12 +30,13 @@ source-repository this type: git - location: https://github.com/eggzilla/ViennaRNAParser/tree/1.2.1 - tag: 1.2.1 + location: https://github.com/eggzilla/ViennaRNAParser/tree/1.2.2 + tag: 1.2.2 library -- Modules exported by the library. exposed-modules: Bio.ViennaRNAParser + Bio.ViennaRNAParserLibrary Bio.RNAcofoldData Bio.RNAcofoldParser Bio.RNAfoldData diff --git a/changelog b/changelog --- a/changelog +++ b/changelog @@ -1,4 +1,8 @@ -*-change-log-*- +1.2.2 Florian Eggenhofer 4. July 2015 + * Added a new module for shared functions between modules + * Sequence and alignment entry parsing is now in line with IUPAC + letter code 1.2.1 Florian Eggenhofer 3. July 2015 * Included parsing of warning field in RNAz output 1.2.0 Florian Eggenhofer 1. July 2015 diff --git a/src/Bio/RNAalifoldParser.hs b/src/Bio/RNAalifoldParser.hs --- a/src/Bio/RNAalifoldParser.hs +++ b/src/Bio/RNAalifoldParser.hs @@ -1,5 +1,5 @@ -- | Parse RNAalifold output --- For more information on RNAalifold consult: <> +-- For more information on RNAalifold consult: < module Bio.RNAalifoldParser ( systemRNAalifold, @@ -9,13 +9,11 @@ ) where import Bio.RNAalifoldData +import Bio.ViennaRNAParserLibrary import Text.ParserCombinators.Parsec import System.Process import System.Exit import qualified Control.Exception.Base as CE - -readDouble :: String -> Double -readDouble = read --- | Run external RNAalifold command and read the output into the corresponding datatype systemRNAalifold :: String -> String -> String -> IO ExitCode diff --git a/src/Bio/RNAcofoldParser.hs b/src/Bio/RNAcofoldParser.hs --- a/src/Bio/RNAcofoldParser.hs +++ b/src/Bio/RNAcofoldParser.hs @@ -9,6 +9,7 @@ ) where import Bio.RNAcofoldData +import Bio.ViennaRNAParserLibrary import Text.ParserCombinators.Parsec import System.Process import System.Exit @@ -17,9 +18,6 @@ -- | Run external RNAcofold command systemRNAcofold :: String -> String -> IO ExitCode systemRNAcofold inputFilePath outputFilePath = system ("RNAcofold <" ++ inputFilePath ++ " >" ++ outputFilePath) - -readDouble :: String -> Double -readDouble = read -- | Parse the consenus of RNAcofold results genParserRNAcofold :: GenParser Char st RNAcofold diff --git a/src/Bio/RNAdistanceParser.hs b/src/Bio/RNAdistanceParser.hs --- a/src/Bio/RNAdistanceParser.hs +++ b/src/Bio/RNAdistanceParser.hs @@ -1,5 +1,5 @@ -- | Parse RNAdistance output --- For more information on RNAdistance consult: <> +-- For more information on RNAdistance consult: < module Bio.RNAdistanceParser ( parseRNAdistance, @@ -8,10 +8,8 @@ ) where import Bio.RNAdistanceData +import Bio.ViennaRNAParserLibrary import Text.ParserCombinators.Parsec - -readInt :: String -> Int -readInt = read -- | Parse the consenus of RNAz results genParserRNAdistance :: GenParser Char st RNAdistance diff --git a/src/Bio/RNAfoldParser.hs b/src/Bio/RNAfoldParser.hs --- a/src/Bio/RNAfoldParser.hs +++ b/src/Bio/RNAfoldParser.hs @@ -10,6 +10,7 @@ import Bio.RNAfoldData import Text.ParserCombinators.Parsec +import Bio.ViennaRNAParserLibrary import System.Process import System.Exit import qualified Control.Exception.Base as CE @@ -17,9 +18,6 @@ -- | Run external RNAfold command and read the output into the corresponding datatype systemRNAfold :: String -> String -> IO ExitCode systemRNAfold inputFilePath outputFilePath = system ("RNAfold --noPS <" ++ inputFilePath ++ " >" ++ outputFilePath) - -readDouble :: String -> Double -readDouble = read -- | Parse the consenus of RNAz results genParserRNAfold :: GenParser Char st RNAfold diff --git a/src/Bio/RNAplexParser.hs b/src/Bio/RNAplexParser.hs --- a/src/Bio/RNAplexParser.hs +++ b/src/Bio/RNAplexParser.hs @@ -8,14 +8,9 @@ ) where import Bio.RNAplexData +import Bio.ViennaRNAParserLibrary import Text.ParserCombinators.Parsec import Control.Monad - -readDouble :: String -> Double -readDouble = read - -readInt :: String -> Int -readInt = read -- | Parse the input as list of RNAplexInteraction datatype parseRNAplexOutput :: GenParser Char st [RNAplexInteraction] diff --git a/src/Bio/RNAupParser.hs b/src/Bio/RNAupParser.hs --- a/src/Bio/RNAupParser.hs +++ b/src/Bio/RNAupParser.hs @@ -8,14 +8,9 @@ ) where import Bio.RNAupData +import Bio.ViennaRNAParserLibrary import Text.ParserCombinators.Parsec import Control.Monad - -readDouble :: String -> Double -readDouble = read - -readInt :: String -> Int -readInt = read -- | Parse the input as list of RNAupInteraction datatype parseRNAupOutput :: GenParser Char st [RNAupInteraction] diff --git a/src/Bio/RNAzParser.hs b/src/Bio/RNAzParser.hs --- a/src/Bio/RNAzParser.hs +++ b/src/Bio/RNAzParser.hs @@ -1,5 +1,3 @@ -{-# LANGUAGE Arrows #-} - -- | Parse RNAz output -- For more information on RNAz consult: Double -readDouble = read - -readInt :: String -> Int -readInt = read - -- | Run external RNAz command and read the output into the corresponding datatype systemRNAz :: String -> String -> IO ExitCode systemRNAz inputFilePath outputFilePath = system ("RNAz " ++ inputFilePath ++ " >" ++ outputFilePath) @@ -102,7 +95,7 @@ _header <- many1 (noneOf "\n") newline notFollowedBy (string (">consensus")) - _resultSequence <- many1 (oneOf "~_-.NATUGCatugc") + _resultSequence <- parseNucleotideAlignmentEntry newline _dotBracket <- many1 (oneOf "-().,") space @@ -128,7 +121,7 @@ parseRNAzConsensus = do string (">consensus") newline - _consensusSequence <- many1 (oneOf "~_-.NATUGCatugc") + _consensusSequence <- parseNucleotideAlignmentEntry newline _dotBracket <- many1 (oneOf "().,") space diff --git a/src/Bio/ViennaRNAParserLibrary.hs b/src/Bio/ViennaRNAParserLibrary.hs new file mode 100644 --- /dev/null +++ b/src/Bio/ViennaRNAParserLibrary.hs @@ -0,0 +1,41 @@ +-- | Library with auxiliary functions needed in multiple other modules. +module Bio.ViennaRNAParserLibrary ( + parseNucleotideSequence, + parseNucleotideAlignmentEntry, + parseProteinSequence, + parseProteinAlignmentEntry, + readInt, + readDouble + ) where + +import Text.ParserCombinators.Parsec + +readDouble :: String -> Double +readDouble = read + +readInt :: String -> Int +readInt = read + +-- | Parse nucleotide sequence. Allowed letters according to IUPAC +parseNucleotideSequence :: GenParser Char st String +parseNucleotideSequence = do + nucleotideSequence <- many1 (oneOf "RYSWKMBDHVNATUGCatugc") + return $ nucleotideSequence + +-- | Parse nucleotide alignment entry. Allowed letters according to IUPAC and commonly used gap characters +parseNucleotideAlignmentEntry :: GenParser Char st String +parseNucleotideAlignmentEntry = do + entry <- many1 (oneOf "~_-.RYSWKMBDHVNATUGCatugc") + return $ entry + +-- | Parse protein amino acid code sequence. Allowed letters according to IUPAC +parseProteinSequence :: GenParser Char st String +parseProteinSequence = do + proteinSequence <- many1 (oneOf "ABCDEFGHIKLMNPQRSTVWXYZabcdefghiklmnpqrstvwxyz") + return $ proteinSequence + +-- | Parse protein amino acid code alignment entry. Allowed letters according to IUPAC and commonly used gap characters +parseProteinAlignmentEntry :: GenParser Char st String +parseProteinAlignmentEntry = do + entry <- many1 (oneOf "~_-.ABCDEFGHIKLMNPQRSTVWXYZabcdefghiklmnpqrstvwxyz") + return $ entry